A familiar workflow.
A smarter starting point.

Make yourself at home.

Download the verified Linux amd64 Debian package. Save it in Downloads, close any running Lamied window, then run:

cd "$HOME/Downloads" &&
sudo apt install "./lamied.ai-0.1.2-amd64.deb" &&
lamied

Use the actual saved folder and filename. The download page also includes checksum verification and reinstall instructions.

Open the right workspace.

Open your project folder in Lamied. Choose the repository root—the folder containing your source and project configuration—so the assistant can work with the right context.

Bring a model you like.

Configure a supported AI provider and model in the editor. Depending on your setup, AI features may require sign-in, provider credentials, and an application subscription. Provider API charges are separate.

You can also configure an OpenAI-compatible local endpoint. Review model and embedding settings before using confidential code.

Give the agent a clear destination.

Describe the outcome, important constraints, and how you want the result checked. The assistant can inspect files, propose changes, and run approved commands.

“Find where order totals are calculated. Replace repeated scans with a lookup map, preserve the existing behavior, and run the relevant tests.”

This is an example prompt, not a command to paste into your terminal.

Read the work. Run the checks.

Watch command output, inspect the changed files, and run your project's tests. Treat complexity findings as guidance, and profile important performance changes with real workloads.

Approve commands individually, or choose remembered permission only for a workspace you trust. You can revoke it later.